Research On User Security Problem Based On Network Traffic Data Mining

In recent years,with the development of computer processing capacity,storage capacity and mobile Internet technology,human behavior has gradually begun to digital direction,which resulted in a lot of data.Data mining,as a powerful analysis tool,can extract hidden useful data and knowledge from massive amounts of data.However,the use of data mining tools for comprehensive analysis of data may violate personal privacy.This article attempts to analyze the potential hazards of data mining to user privacy.In order to better and more accurately mine the user's deep information,we need to do a good job of information collection.This paper comes up with a model of traffic classification,Firstly,we extracted the shallow information in the flow of user's accessing Internet,which contains data acquisition,data cleaning and data transforming.then we extracts the user behavior characteristics through data mining algorithms such as pattern mining and association rule mining,Finally we give the pattern matching algorithm,the unknown sample flow to extract the pattern and the existing model library to match to achieve classification;based on these steps,we can get enough valuable data to analyze the user's deep information.This paper explores two types of deep information: estimating user consumption and predicting user behavior sequence.Estimating user consumption is based on information association and statistical analysis,and the use of Bayesian formula to obtain the probability of buying all kind of service or goods.The user behavior sequence prediction is based on the Markov prediction model.Finally,we verify the effectiveness of the classification model and prediction model through implementation of the model and doing a series of experiments on the real access data in our labs.Which proves that in the large data environment,through the data mining association analysis and other means,can be a great impact on user privacy and security,such as estimating the user's consumption situation,predicting user's actions in the future,the same way could also be used to get the user's gender,age,occupation and so on.Consequently,with the big data environment,the user's privacy and security issues need to cause widespread concern.